Influence of digital elevation models on landslide susceptibility with Logistic Regression Model

• Autor(a) principal: Gonçalves, José
• Data de Publicação: 2018
• Outros Autores: Faria, Ana, Bateira, Carlos, Fernandes, Joana, Oliveira, Ana
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10451/39052
• topic: Statistical Modelling; Landslides; Agriculture Terraces; Douro Demarcated Region

This paper focuses on the influence of Digital Elevation Models on the landslides susceptibility assessment in agricultural terraces, using Logistic Regression statistical model.This study was performed in a watershed located at Carvalhas Estate in Douro Valley, using an inventory of 109 landslides. To analyse the influence of the digital elevation model (DEM) resolution we used three DEMs, (A), (B) and (C). The DEMs (A) and (B) were directly obtained by processing aerial images and extracting different resolutions, 1 and 5 meters, respectively. The DEM (C), with 5m resolution, was processed with Topo to Raster interpolation method, using as input data contour lines of 10 m interval, elevation points and hydrography. The Logistic Regression was performed using two models which are distinguished by the independent variables selection. At model 1 was used the slope, curvature, raiser slope, riser height, contributing areas and topographic wetness index. In model 2 we decide remove the independent variables related with the terrace geometry, riser slope and riser height. The result seems to indicate that there is no significant influence of different resolutions of Digital Elevation Models in susceptibility modelling at this small scale and using statistical methods. The independent variables riser slope and riser height provide information of the terraces geometry and the construction techniques that enter the modelling process with more detailed information.